SkillWard Agent Skill Security Scanner
Verified by a Dify partnerScan AI Agent Skills and MCP projects before installation or deployment. This template lets users provide one scan source, such as an uploaded .zip package, a Git repository URL, or a remote MCP URL. The SkillWard plugin sends the source to the hosted SkillWard Sentinel backend for static analysis, LLM-based safety evaluation, and optional runtime sandbox verification, then returns a concise security summary and structured scan report.
ITKnowledgeOthers
Setup steps
- Install the SkillWard plugin from the Dify Plugin Marketplace.
- Authorize SkillWard in Dify by navigating to Tools → SkillWard → Authorize.
- Configure the required OpenAI-compatible LLM credentials: API key, API base URL, and model name. API version is optional for providers that require it.
- In the User Input (Start) node, let the user provide exactly one scan source: an uploaded .zip Skill/MCP package, a Git repository URL, or a remote MCP URL.
- If the user uploads a file, map it to the SkillWard tool node’s archive_file input. If the user provides a URL, map it to target_url and select url_type as git_repo or remote_mcp.
- Configure the scan options in the SkillWard tool node: enable LLM evaluation, runtime sandbox verification, tool-chain analysis, and select the report language.
- The SkillWard tool node sends the source and scan options to the hosted SkillWard Sentinel backend. The Sentinel endpoint is built into the plugin and does not need to be configured separately.
- The Answer node displays the scan summary and key findings from the SkillWard security report to the user.
TYPE
Workflow
CATEGORY
it
knowledge
others
PLUGINS
Dify Compatibility
1.13.0
SUPPORTED LANGUAGES
English
LAST UPDATED
Jul 14, 2026
Give a rate
SHARE WITH